In the Naturales Quaestiones, a collection of scientific reflections and short treatises on natural phenomena, Seneca subordinated the study of physics and nature to ethics, which he considered the ultimate purpose. A deep knowledge and understanding of the rules that govern the world was, for the philosopher, a necessary condition to analyze its mechanisms and phenomena.
You might wonder what Seneca has to do with technology and artificial intelligence. The answer is: the mirror.
Seneca dedicates the first book of this work to celestial phenomena, and in particular to those—such as the rainbow—that represent imperfect reflections of a celestial body. In this way, the author can reflect on those artificial mirrors that do not faithfully reproduce the image, but may distort it, show it askew, or even amplify it.
Nature, Seneca writes, has provided us with a tool for knowledge through the mirror—understood as any surface that reflects real forms, from water to polished stone. The mirror is essential for humans, he argues, because it allows us to observe the sun—otherwise unbearable to look at directly—and, crucially, to know ourselves.
Â
From a philosophical perspective, the mirror becomes then a perfect instrument for self-knowledge and discovery.
Â
 This is the image that first came to mind some time ago, when I found myself looking at a set of numbers describing a risk classification in a sensitive context, generated by an artificial intelligence system. I thought of AI as a mirror—one that was distorting and amplifying deeply human, all too human problems.
In that case, the system was assigning risk scores to individuals based on patterns extracted from historical data. At first glance, the output appeared technical, neutral, even objective. But looking more closely, what it revealed was something else entirely. It was reflecting existing inequalities and biases embedded in past decisions. It was rendering visible, in numerical form, assumptions that had long operated implicitly. Paradoxically, when scrutinized and assessed, it revealed the problematic nature of the decisions it helped to reach: like a mirror, it reflected a distortion that had been lost over time—one that no human being would ever have admitted with such bluntness. It can expose structures that remain invisible in everyday decision-making, precisely because they are embedded in habits, institutions, and histories. We know that AI systems do not simply represent reality. They transform it in the very act of reflecting it, by selecting, compressing, weighting, and operationalizing certain aspects of the world while ignoring others. In doing so, they can make some patterns more visible, more actionable—and therefore more powerful.Â
Artificial intelligence, therefore, can help us reflect on our own limitations and expose certain patterns that are primarily about ourselves. But like Seneca’s mirrors, they do not necessarily return a faithful image, but one shaped by the surface, the reality that produces it.
This is why the mirror is at the centre of Immanence website’s new visual identity, as a tool of observation and reflection. It considers technological systems as a reflection of who we are as individuals, societies, and organizations—in both positive and negative ways: they can reveal—as in the example above—what we would prefer to hide by using them (such as our responsibilities) or they can help us better express ourselves and our values.Â
Â
The choice is ours, and that is precisely why we are here.Â
By the end of the year, the goal was to bring artificial intelligence out of the cloud—that is, to stop thinking of these technologies as something abstract, hovering above our heads, neutral, and inevitable. It means bringing them back into real-world contexts: within organizations, within infrastructure, among people, and into decisions that produce concrete consequences.Â
In philosophy, immanence refers precisely to this. It is a concept born in opposition, particularly to that of transcendence. It denotes what exists as part of reality; it denotes interaction with the subject and the context, as an inseparable whole. It is a characteristic inherent in everything that does not transcend—that is, that does not refer to something “superior”. This is also how we like to think of artificial intelligence.
In fact, Immanence is celebrating its third anniversary and we are launching a new website that directly responds to this concept. Technologies exist in society within a relationship of absolute reciprocity. They do not merely produce effects on the world: they are also its product, reflected in a mirror. They arise within specific material conditions, within visions, within economic and organizational systems. And, once introduced into social contexts, they begin to transform them in turn.Â
In her book The AI Mirror: How to Reclaim Our Humanity in an Age of Machine Thinking, Prof. Shannon Vallor draws on the analogy between AI and a mirror to identify where we are going wrong in our relationship with this particular technology, and recounts an episode from 2022 involving Google engineer Blake Lemoine, who claimed that a Google language model had become sentient and had therefore developed the capacity for independent thought and action. Vallor argues that Lemoine was actually more or less talking to himself, and thus leads the reader to a fourth theme associated with mirrors: that they create the illusion that there is something on the other side, which in reality does not exist separately from what is reflected.
In the past (and often still today), humans were guided by mirroring in the constellations: we projected stories, meanings, and directions onto distant points of light, using them to navigate both physically and symbolically. The sky did not speak, but we learned to read it. We recognized patterns, even where none had been intentionally placed, and in doing so, we constructed systems of orientation.
Something similar happens with artificial intelligence.
We look at these systems and search for intention, agency, even understanding. We interpret coherence as thought, fluency as awareness, responsiveness as presence. And sometimes, like in Lemoine’s case, we begin to believe that there is something on the other side – a subject, separate from us, looking back. But the mirror does not contain another subject, yet a reflection.
This makes it not easier, but actually more demanding. Because if what we are seeing is not an independent intelligence, but a reflection shaped by data, design, and context, then the responsibility does not lie with the system. It lies with us—those who build it, deploy it, integrate it, and decide what to do with what it shows.
What are we seeing? What are we amplifying? What are we choosing to act upon? What are we choosing not to see?
Artificial intelligence does not remove the need for judgment. If anything, it intensifies it. It produces reflections that can guide decisions, but also mislead them; that can clarify patterns, but also entrench them. Like Seneca’s mirror, it can help us observe what would otherwise be unbearable to look at directly. But it does not absolve us from deciding what to do with that knowledge.
This is, perhaps, where the idea of immanence returns in its most concrete form. To keep artificial intelligence within the world – within our institutions, our responsibilities, our systems of accountability – means refusing the temptation to treat it as something external, something that decides for us, something that can be blamed in our place. This, I’m sure, will force us to look within ourselves.
Â
Â
Fonti:Â
Sabine Melchior-Bonnet, Storia dello Specchio, Bari, Dedalo, 2002.
Alessandro Moscone, Lo specchio e il saggio: tra virtĂą e vizio, Treccani, 2024.
